Unitary representations of reductive Lie groups (Q957703): Difference between revisions

This paper is based on the lecture given in the conference [Mathematics towards the third millennium. Papers of the international congress, Rome, Italy, May 27--29, 1999. Atti Accad. Naz. Lincei, Cl. Sci. Fis. Mat. Nat., IX. Ser., Rend. Lincei, Mat. Appl. 2000, Suppl., 7-167 (2000; Zbl 1151.00308)]. In the first one third, the author describes the Gelfand program, which tells us how to use ``representation theory'' to get various properties of the space \( X \) on which a certain group \( G \) acts. Two examples are given. One is a ``toy model'' of Radon transform on a homogeneous space of the symmetric group of order \( n \). The other is the cohomology of locally symmetric spaces based on his joint paper with Zuckerman [\textit{D. A. Vogan, Jr., G. J. Zuckerman}, Compos. Math. 53, 51-90 (1984; Zbl 0692.22008)]. There is a brief account on the proof of Matsushima's formula, which illustrates the role of representation theory. This part is one of most pleasant readings for beginners of representation theory, and it bears a good atmosphere of the lecture. The second part is devoted to explaining orbit method and the geometric quantization in the spirit of Kirillov-Kostant. The orbit method roughly suggests that, in order to classify unitary representations of \( G \), one should attach each irreducible unitary representation to a coadjoint orbit in the dual of the Lie algebra of \( G \). The reversed direction, from orbits to representations, are provided by the geometric quantization. If \( G \) is reductive, there are three important class of coadjoint orbits, namely, hyperbolic orbits, elliptic orbits and nilpotent orbits. The rest of orbits are in some sense a mixture of these three, and once if one can understand representations attached to these orbits, then the construction of the representations (\(=\) quantization on arbitrary orbits) should be well understood. For a hyperbolic semisimple orbit, there corresponds a representation which is induced from real parabolic subgroup. Similarly, for an elliptic orbit, cohomological induction from \( \theta \)-stable parabolic provides a representation attached to the orbit. However, up to now, there is no systematic way to produce a representation for a nilpotent orbit. Such representations attached to nilpotent orbits are often called ``unipotent''. In the last part of the article, the author explained a possible strategy to construct unipotent representations and to prove their unitarity. The key roles are played by the Kazhdan-Lusztig theory, certain bound of infinitesimal characters and the multiplicity of characteristic cycles. For this, an example for the split real group of type \( \mathrm{G}_2 \) is given.